Galaxies are complex systems of stars, gas, dust and dark matter. The processes regulating galaxy evolution are reflected in clear correlations between galaxy properties, also called scaling relations. It is only recently (thanks to modern space telescopes) that scaling relations involving dust properties could be investigated. While several interesting links between dust and stellar properties are found, it is not clear what the main drivers of these relations are. One of the reasons for this is the lack of a self-consistent model, describing the local interplay of stars and dust, and its effect on the galaxy's emission. We propose to build a modelling tool based on simulations of stellar light travelling through a dusty medium. These simulations will be designed to mimic galaxies in a realistic, 3-D way and properly treat absorption, scattering and thermal reemission by dust grains. We intend to set up an extensive and intelligently designed grid of galaxy models and compute their spectral energy distribution. This library will subsequently be used to fit the observed emission of galaxies from the UV to the sub-millimeter range. We will apply this tool to over 900 local galaxies in an unprecedented systematic study of the dust scaling relations and their physical origin. In a later phase, this study can be extended towards higher redshift to probe the evolution of galaxies through cosmic time.